Impedance Is, Well, Complex...

[Grainger49]

There are all sorts of questions about speaker impedance on the board.  It is generalized by manufacturers as a "nominal" number.  But that doesn't describe much of anything.  The two charts below are from the Fuselier 3.8D speakers I had when I started on my Bottlehead journey.  They were described as a nominal 8 ohms impedance.  As you can see they are 8 ohms at 5 different frequencies and not 8 ohms everywhere else.  They were not SET friendly:

The top graph is the kind that Stereophyle publishes.  It shows the magnitude of the impedance only.  The problem is that impedance is a complex number.  There is a resistive and reactive component to it at every frequency.  So the second graph shows a "Polar" impedance curve.  The X axis is resistance the Y axis is reactance.  Anything above the X axis is inductive, anything below is capacitive.  Anything to the right of the Y axis is a positive resistance, to the left of the Y axis is a negative resistance. 

See is it complex.  And never mind the square root of -1 thing that is buried in this too.

I wish I knew what the loops in the curve mean.  It might have to do with the phase correct crossovers John Fuselier was known for.  I can only tell you that these speakers were soundstage champs.  But they needed to be driven by more power than is available from SET amps.

So, a little about why a single number doesn't describe impedance.

 

[[email protected]]

Grainger, I think you just wrote part of my lesson plan for Trig on Monday.  We just started our section on polar coordinates on the complex plain.  I've got eleven kids in that class and every one is looking at me like "Who on earth ever uses this stuff?".  You just answered the big question.

What is really interesting is that the second graph is actually a function.  The vertical line test cannot be used on the complex plane.  You have to show the input values along the line because each point on the complex plane is just one number (not a pair like the cartesian plane).

You are dead on.  Impedance is "complex".

 

[Grainger49]

Whenever you talk about alternating current you have complex numbers, X plus or minus i Y, or a vector with magnitude and an angle in standard rotation.  Everything you do goes polar to rectangular and back.

 

[RPMac]

I have EV SP12B and SP8B "16 ohm" speakers. If I parallel these to get an "8ohm" load to connect to my "8 ohm" EXO-36, would this give me a more "tube friendly" across frequency impedance load?

 

[Paul Birkeland]

If you don't need the extra power, 16 Ohm speakers on an 8 Ohm speaker tab can be a very nice setup with a single ended tube amp (you trade a little power for a little less distortion and greater damping).

If you want to parallel two 16 Ohm speakers that are different, it won't hurt your average single ended amp to try.

-PB

 

[Grainger49]

Robert,

I have been out of town/off line for a few days.  It is going to get worse starting next week.  I am employed again!

I am of the opinion that you should try the available taps and see which sounds the best to you.  I mean, whose ears are you trying to please anyway?

Let's think out loud, so to speak or type... Paralleled resistances are always lower, if equal halved.  So the impedance peaks are cut in half, that is a good thing.  BUT the impedance dips are also doubled (or cut in half, depending on how you look at that).  At the very least you should look at an impedance magnitude graph (like the top one above) to see that the driver doesn't dip to 2 ohms somewhere... like the Fuselier 3.8Ds do at 10k Hz.  That would be a one ohm impedance to the amp. 

BTW, I owned serial numbers 3 and 4 of that speaker, made for me.  Serial numbers 1 and 2 were at Fat Julian's Audio shop.  John Fuselier was a customer when I was a salesman and a friend there after.  A really great guy, he would fit in with the Bottlehead crowd.